15:152:1522:1522

15.2.2 Damages due to Problems in Forged Parts and Powder-Metallurgical Parts

Forged materials are still dominant today in parts under high cyclical stress, especially rotors (LCF) and compressor blades (HCF). The ever greater loads (centrifugal forces, thermal stress) demand reliably safe prevention of ever smaller flaws. The design process now incorporates material properties that make it possible to determine the crack initiation near flaws, crack growth, and the critical crack length at which fracture occurs. This information is specified directly as material data and indirectly through the material structure. Therefore, unallowable deviation of the structure must be defined as a damage-relevant flaw, just as a crack or dangerous segregation would be.
For many years, there have been repeated reports regarding titanium-specific material problems in large forged parts in rotors (Ills. 15.2-13, 15.2-14, 15.2-15, 15.2-16, 15.2-17, 15.2-18). In spite of intensive efforts to correct these problems, they still seem to occur to a certain degree.
A spectacular damage incident (Volume 3, Ill. 14-10) related to a powder-metallurgic material (PM) with high-temperature strength has had a sobering effect following initial euphoria regarding these materials. This type of incident can delay the serial implementation of a technology by decades. The utilization of the high strength of these materials made them especially sensitive to process-specific flaws (Ill. 15.1-16).

Illustration 15.2-11 (Ref. 15.2-15): This diagram depicts the most common flaw types in forged parts with their effects on the part properties. It also includes any relevant references to sections that deal with the specific problems in greater detail.

“Grain structure”: This term is confusing. In materials of “engine quality”, it merely refers to structural orientation. This is primarily determined by a favorable grain boundary pattern and the location of carbides in the grain.
Unlike materials such as construction steel, which have a considerable proportion of material-typical inhomogeneities (not flaws!), the high-strength materials used in engine construction are especially pure. However, even here there is an influence of grain boundary patterns (e.g. under creep loads, plastic deformability) and any inhomogeneities that affect strength (weak points). One attempts to minimize negative influences, especially those that affect dynamic fatigue and creep life, through favorable grain structure orientation (texture), i.e. parallel to the highest operating loads. This orientation is achieved with the aid of plastic deformation during the forging process (Ill. 15.1-13). It also corrects flaws such as unallowable segregations. However, the altered location of these flaws can make them more difficult to detect with ultrasonic testing (Ills. 15.2-21 and 17.3-15). Therefore, non-destructive testing must be optimized accordingly. It is not always possible to guarantee sufficient forged re-forming, especially in thick cross-sections such as the hub area of a disk. This promotes dangerous flaws precisely in this especially highly stressed part zone. This includes all flaws from melting (Ill. 15.1-12) and casting porosity that hasn`t been completely forged. An additional effect is the smashing of larger inclusions into many smaller, less dangerous small ones (Ill. 15.1-16). Therefore, unfavorable grain structure and/or insufficient forging deformation increase the risk of life-reducing flaws. The cyclical life span (LCF) is especially affected.

Crack initiation: “Forging cracks” usually occur near the surface. They are caused by friction forces between tools (forging hammer, forging die, extrusion die) and the part being forged. Internal forging cracks that are not oxidized will usually disappear by being welded shut during the forging process. Common crack locations are in the hub area. Many forging cracks in the surface zone of finished parts can be avoided by ensuring that there is a sufficient oversize in the raw part. The remainder should be sufficiently safely detectable through serially implementable non-destructive testing methods (e.g. ultrasonic, eddy current, and penetrant) that have been optimized for the specific parts. The trend toward ever higher utilization of material strength makes the limits of detection for these processes even tighter (Ill. 17.3.1-1).
Heat treatment cracks (Ills. 15.1-19 and 15.2-13) occur as a result of overly high thermal stress and insufficient strength in the heating or cooling phases. If the cracks oxidize or if the deformation is insufficient for them to reseal during the forging process, these cracks will not be “healed”.

Deformation-dependant strength losses occur when sufficient deformation does not occur during forging. This is especially common in thick hub cross-sections that have a high resistance to solid forging. This means that the required static and dynamic minimum strength values may not be met in local areas. The reason for this is less than optimal development of the structure (Ill. 15.1-4). Especially affected materials are ones in which the strength must be ensured through a strictly specified thermomechanical reforming process. It primarily affects LCF strength and creep resistance.

Strength losses through unfavorable temperature control: the larger the part cross section, the slower temperature changes will occur (Ill. 15.1-14). This is especially true of titanium alloys with their low thermal conductivity. However, even in raw disk parts made from Ni alloys with large masses in the hub area, as are typical for turbine disks of small gas turbines (middle diagram), there is a danger that the structure will not be optimal (Ill. 15.2-12).

Residual stresses are created in parts during the forging process through deformation and/or restricted thermal strain during subsequent heat treatment. If tension residual stresses occur in a part area that is subject to high operating loads, they can unallowably shorten the life span of the part (Ill. 15.2-19). The greater the thermal strength of the part, the greater the residual stresses that cannot be broken down through stress-relieved annealing, and these can combine with the operating stresses in the part (Ill. 15.2-19).

Thermomechanical Processing (TMP) can create structural zones that appear light during etching and are referred to as “pseudo white spots” in the technical literature (Ref. 15.1-4). However, these are not inhomogeneities of alloy components that present a cause for concern, but are rather coarse grain zones (detail in bottom frame) that are surrounded by very fine grain. These coarse grain zones are created when the annealing temperature does not completely reach the level required for complete recrystallization. The difference may only be a few °C.
This phenomenon is especially known from the material IN 718, which has a broad temperature range between 995°C, where crystallization begins, and 1030°C, at which point the entire structure has been reached.
The recrystalization model (bottom frame) uses a structured angled plane to show how the individual structural zones recrystallize when they reach the necessary energy state (temperature).

Illustration 15.2-12 (Ref. 15.2-5): the development of the g' phase, which is responsible for thermal strength, depends largely on its solution and separation during cooling. In this process, the temperature levels during heat treatment are decisive. Especially important are the heating and cooling rates in the entire part, as well as strict adherence to the solution annealing temperature. Even small deviations from the solution annealing temperature of only 2% can, for example, change the important g' component by over 20% in the depicted high-strength Ni forged alloy, which has a decisive influence on thermal strength.

Illustration 15.2-13: During a testing rig run in the development phase of a military engine with multiple shafts, a serious imbalance occurred. Disassembly revealed a gaping circumferential crack about 20 centimeters long on a turbine disk at the flange connection to the disk (right diagram). Unlike the disk surface, the open crack surface was heavily oxidized. A metallographic cut confirmed the findings (top detail). This oxidation could not be explained with the known operating conditions. A fracture surface analysis did not show any dynamic fatigue crack signs, except for the ends of the crack. The crack surface, as far as it could be analyzed, revealed the typical “doughy” characteristics of a thermal crack (Ill.15.1-8). Further research revealed that the crack occurred during heat treatment of the part in its raw part contour, and also occurred in several additional raw parts (left diagram). The raw part contour had a notch and abrupt cross-section jump to a rapidly heating thin ring in the area where the cracking later occurred. This allowed dangerously high thermal stress to develop during heating and cooling. It is not known why the part was finished and installed in the engine despite the large crack length (Ill. 17.3.1-9). The decisive factor may have been the “horizon of expectation” of the crack testers, which was geared to much smaller cracks, combined with the oxides filling the crack.

Illustration 15.2-14 (Ref. 15.2-30): The Sioux City flight accident (Ill. 15.2-18) was traced back to a material flaw in the fan disk of one of the three engines. The material was a titanium alloy. A review team made up of specialists was tasked with evaluation of the entire technology, including raw part production, construction, and design configuration. They were also to determine possible suggestions for improvements. In this context, 22 relevant damages and incidents that occurred before 1990 were compiled and analyzed.
The causal material flaws could be classified into four categories corresponding to their microstructure, chemical analysis, and the physical characteristics that influenced the fracture surfaces:

Type I/Category 1: This flaw is relatively common and is related to process controls and quality problems. It occurs due to “burned” (reaction with air) titanium sponge from the melting process. The typical result is three defined concentric zones (top left diagram) Zone 1 in the inside has a sponge-like structure with large pores with an average diameter of 0.75 mm. Zone 2, which only occurs in this flaw type, consists of very hard and probably also brittle nitrogen-stabilized a structure. Zone 3 is the same in all four categories. It is an ellipsoid of a structure, with its orientation corresponding to the forging deformation, and surrounded by a-accumulations or a-plates.

Type I/Category 2: This flaw type is also relatively common and is related to process controls and quality problems. Zone 1 can contain pores, surrounded by a zone 3 made of nitrogen-stabilized a structure. The hardness is RC 55-70, which is clearly higher than the normal RC 35-40. The pores have an average diameter of 0.1 mm, making them considerably smaller than in Category 1. Pores of such a small size evidently do not explain the crack growth in damage cases. Apparently, the entire flaw zone has a weakening effect. Here, as well, zone 3 consists of nitrogen-stabilized a structure.

Type II/Category 3: Is considerably rarer than categories 1 and 2. It consists of a zone 1 with very small micropores with diameters of between 0.25-0.005 mm. Here, as well, zone 3 is made up of a structure, although in this case is it aliminum-stabilized. The hardness is RC 35-40, thus corresponding to the normal values of the base material.

Type II/Category 4: These flaws have been the least common as damage causes. They consist merely of a single zone 3. These segregations consist of pure titanium and have a correspondingly low hardness/strength (RC 12). Porosity has not been found. Flaws of this type were found before the introduction of “hot topping”.

Crack growth in the cracking zone in the investigated damage cases showed the following typical characteristics:

  • crack initiation at pores
  • smooth cracked surfaces (cleavage cracks) directly around the pores correspond to brittle micro-cracking (Ill. 15.2-20).
  • inter-crystalline, facet-like cracking along plate-shaped a structures.
  • LCF cracking and ductile splitting between the facets.

Depending on the flaw type, the fracture surface characteristics had varying degrees of pronouncement. In the case of category 1, porosity evidently had the dominant influence on crack growth. In categories 2 and 3, a similar structural influence could be discerned. An important realization was the fact that the pore size does not necessarily correspond to the effective crack initiation flaw size. For this reason, in the case of risk assessments, the assumption of a larger effective pore size seems to be necessary (Ill. 15.2-14.1).

Illustration 15.2-14.1

Illustration 15.2-15 (Ref. 15.2-30): The analysis of flaws in rotor parts made from a titanium alloy in the time period before 1990 resulted in the following conclusion: the frequency of flaws from the melting process (Ill. 15.1-12) and the forging process (Ills. 15.3-12 and 15.3-13) is not equal throughout the entire part volume. One can see that dangerous flaws occurred, or at least were discovered, especially in the hub area. This type of flaw distribution is also valid for Ni-based alloys and high-alloy steels produced in similar processes. Flaws in the hub area are especially relevant to safety. This area is typically subject to very powerful cyclical loads. In addition, in the thick hub cross-sections, it is more difficult to realize the plastic deformations during forging that are required to achieve the necessary strength properties (Ills. 15.1-14 and 15.2-11). This deformation is a prerequisite for the smashing of flaws (impurities, also see Ill. 15.1-16) and/or giving them a favorable orientation relative to the main load direction (Ill. 15.1-13). If the deformation is not sufficient, larger flaws in more effective positions are more likely.
About 80% of the burst rotors had crack initiation weak points below the surface. The LCF cracks grew towards the surface. This meant that dangerously large cracks could only be detected at an advanced stage through penetrant testing. Of course, it was a prerequisite that this type of testing occurred during this growth phase of the cracks, such as during overhauls.
Generally, it can be said that the probability of material flaws increases along with the volume of the parts, i.e with the thickness of the cross-sections.

Illustration 15.2-16 (Ref. 15.2-18, Example 15.2-3): The one-piece compressor rotor (spool) made from the high-strength titanium alloy Ti6242 failed between the third and ninth stages (middle right diagram) due to LCF (bottom left diagram). The crack was traced back to a decrease in dynamic strength in the area of an oxygen accumulation (segregation, bottom right diagram) with an increased proportion of a-structure. The flaw area had a slightly higher hardness of RC. 38-43 relative to the matrix hardness of RC 34. This was most likely a Type II, Category 3 flaw (Ill. 15.2-14). The weak point was near the highly-stressed circumferential dove-tail groove.
The oxygen accumulation occured during the triple remelting process. A possible cause was a pronounced vacuum loss during the second remelting process. This occurrence could be reconstructed with the aid of the required documented process records (Ill. 15.3-10). It is rare for such a process to occur with unallowable intensity. It occurs when the electrode shifts to a cooling water leak in the mold. However, the pressure increase was still within the limits that were tolerable at the time the raw material was produced (1972). The oxygen from the water, which dissociated at the high temperature, then diffuses into the melt bath (Ill. 15.2-17).
Typical characteristics of such a procedural deviation were attributed to the damaged part. This made it possible to identify other potentially threatened parts that were located near the damaged part in the casting block (also see Ill.15.2-22). In compliance with the demands of the responsible authorities, all suspect parts (21) were removed within 30 days.
This type of weak point with no cracking could evidently not be found through ultrasonic testing due to its unique location. The flaw, insofar as it spread to the surface, would have been detectable with the aid of macro-etching (blue etch anodizing = BEA). However, this process was not yet in use at the time.
Even in the finished part with potential cracking, ultrasonic testing during an overhaul or inspection was problematic. The flaws in the groove area were in a location that was most unfavorable for ultrasonic testing (blind spot). This is especially true of cracks that are located directly below the surface. During the last overhaul, ultrasonic testing showed anomalous findings, but they were categorized as allowable in accordance with regulations.
For this reason, additional eddy-current testing was recommended for the detection of cracks near the surface.

Example 15.2-3 (Ill. 15.2-16, Ref.15.2-18)

Excerpt: “…Shortly after the commencement of the take-off roll, at about 20 knots, there was a loud explosion and the aircraft yawed sharply to the left. The takeoff was rejected, and there was a fire warning on the left engine…
Approximately 30 kilograms of rotating hardware from the left engine HPC and the compressor case was found on the ground near the aircraft. No engine debris penetrated the passenger cabin…
The engine's inlet gearbox was fractured, causing a disconnect of the engine accessory drive, which includes the main engine fuel pump…
The uncontained failure of the third stage of the 3-9 high-pressure compressor spool was due to the presence of an oxygen rich segregate…“

Comments: Several similar damages in the same engine types/rotor design have been reported over a longer time period. Evidently, in all cases, LCF cracks originated in material flaws following cyclical loads from the startup/shutdown cycles. The problems seem to lie in the quality assurance of the raw part production for large titanium rotors with thick cross-sections. As far as one can tell, this type of problem cannot be completely ruled out in other engine types even today.

Illustration 15.2-17 (Ref. 15.2-18): The optimization of the manufacturing process for aviation-suitable raw parts made from high-strength titanium alloys takes years. Today, remelting is usually repeated three times in a vacuum. The following describes typical process steps:

In step 1, pure titanium sponge is mixed with the powdered alloy components. This is pressed into large cuboids which are welded under shielding gas with strips into the first electrode with a diameter of 45cm (measurements of typical example; left diagram). An intermediate piece connects the electrode with the header.
This electrode is melted by an arc into a water-cooled mold, where it becomes a second electrode with a diameter of 75 cm.(see Ills. 15.1-11 and 15.1-12). Because material can melt from the intermediate piece during this process, it must also be made from rotor-quality titanium, as must the connecting straps.
For the second remelting, three of the ingots created in step 1, in this case with a diameter of 60 cm, are combined into an electrode (middle diagram). In the second remelting (step 2), unlike in step 1, no head material may be melted. A melting process similar to that in step 1 results in an ingot with a diameter of 75 cm.

The resulting ingot is inverted and creates the electrode for the third remelting (step 3). This creates the final product of the melting process, which is an ingot with 90 cm diameter.
This ingot is turned, heat-treated, and forged into billets (see Ill. 15.3-11). It is then split into lengths that are required for the specific raw part volume. Machining creates the geometry necessary for the forming process into part-specific blanks (e.g. drop-forging part).

Notes for Ill. 15.2-16: Originally, the blanks for rotors in stages 3-9 (middle right diagram) were made from a single pre-forged billet with a diameter of 40 cm. Then, billets with diameters of 30 and 32.5 cm were used. Later, two 25 cm billets were used, and these were eventually replaced by 20 cm billets. The two parts were prefabricated and then welded into a rotor. The major re-forming of the thin billets into a forging blank leads to a more favorable micro-structure, and reduces flaw sizes and changes their orientation. The trend towards ever smaller billet diameters was also intended to aid detection of hard a segregations. It created a larger surface that could be inspected through etching/BEA (blue etch anodizing).

Illustration 15.2-18 (Ref. 15.2-30, Example 15.2-4): After 41,009 operating hours with 15,503 startup/shutdown cycles, an LCF fracture occurred in the fan disk (middle diagram), which was made of the titanium alloy TiAl6V4. 760 cycles before the damage, the engine was in for repair, in the course of which the disk was inspected with fluorescent penetrant. An ultrasonic inspection of the dovetail grooves in the annulus was also conducted. No cracks were reported by these inspections.
The crack that led to the fracture originated in a 1.4 x 0.3 mm flaw (left bottom detail) near the hub bore. The cyclical (LCF) crack growth between 14,000 and 16,000 load cycles led to the critical crack length of roughly 30 mm at the surface of the hub bore. In the crack initiation zone, there was an accumulation of hard alpha material with a high nitrogen content. In this zone, there were also microcracks and microporosity. These are typical characteristics of flaw type I (Ill. 15.2-14). In a “sister disk” from the same melting batch, ultrasonic inspection revealed anomalies in the disk membrane of the hub bore and the flange arm. They were the same flaw type as in the damaged disk. However, there was no LCF crack growth. In addition, macro-etching (blue etch anodizing) showed segregations that could be categorized as flaw type II with a high aluminum content.
The flaws with the increased nitrogen content were evidently brought in during the remelting of the ingot (Ill. 15.2-30).

Example 15.2-4 (Ill. 15.2-18)

Excerpt 1 (Ref. 15.2-19): ”…(The aircraft) that encountered a critical engine malfunction and complete loss of its hydraulic systems…crashed and burned…while attempting to land at a municipal airport…More than 100 people died in the accident…“

Excerpt 2 (Ref. 15.2-20): “The first stage disk from the No. 2 engine…that crash-landed…at Sioux City…has been recovered and last week became the immediate focus of intense inspection. Preliminary investigation showed a “pre- existing crack” on the interior surface of the disk…“

Excerpt 3 (Ref. 15.2-21): The accident disk also bore evidence of a preexisting fatigue crack zone near its inside surface where the disk encounters its highest stress….How such a crack could go undetected is the focus of an investigation…Inspectors also have discovered a 0.055 x 0.012-in (1.4 x 0.3 mm) cavity on the accident disk's surface that may be evidence of a hard alpha but may not have been detected with dyes. The difficulties of detecting cracks may require research into new inspection technology…“

Comments: This case is the well-known Sioux City accident. The problems of material flaws in large rotor disks, especially those made from high-strength titanium alloys, is still an important topic roughly 15 years later. Through improvements, especially in the melting process and in nondestructive testing, safety was improved. However, similar damages continue to be reported (Ill. 15.2-16). The top illustration is intended to give an impression of the size of the main fragment that was later found in a field.

Example 15.2-5:

Excerpt 1 (Ref. 15.2-22): “The first incident occurred when the fan of a 250-cycle (flight)…from the No. 3 position…(from an Aircraft) taking off from New York fell into the sea.
…the second incident occurred…en route from Los Angeles to Chicago (a 300-cycle fan which was not in the same batch). …(in this) accident the No. 1 engine fan left the pod, travelled foreward under the fuselage and then aft, striking the No. 3 Nacelle.”

Excerpt 2 (Ref. 15.2-17): “Within a short time it was already determined (through inspections) that the A disks from the top half of the… titanium ingot were more prone to damage than the B disks from the bottom half. Before the investigation, (the OEM) assumed that there was no difference between these disks, but a statistical analysis revealed that cracks only occurred in the A disks. In order to minimize the residual stress, (the OEM) increased the wall thickness of the disk.”

Comments: Although no fragments were found in either case, it can be assumed that a causal influence was evidently high tension residual stresses from the production process of the raw parts. It is likely that pores and structural flaws from the remelting process contributed to the damage. In addition, the LCF life-reducing influence of dwell time was underestimated (Ref. 15.2-16).

Illustration 15.2-19 (Ref. 15.2-17, Example 15.2-5): Disk fractures occurred in a large first-generation fan engine. The material was a high-strength titanium alloy. Operating cracks (LCF) in several other disks that were subsequently inspected were causallly attributed to problems during raw part production. Because the fragments could not be recovered, the inspection results of the parts with operating cracks are treated as representative of the accident cases. This means that these cracks are assumed to correspond to the damage mechanism of the disk fractures.
Evidently, tension residual stresses from the forging process (bottom right diagram) overlayed with high operating loads in the affected disk area.
In addition, embrittling structural flaws with microcracks and microporosity (Ill. 15.2-14) were present. They originated in the upper cast bar zone (Ill. 15.2-17) and remained in the pre-forged material (A-billets, bottom left diagram). The LCF crack probably originated in this type of flaw.

Illustration 15.2-20 (Ref. 15.2-17): Pores in titanium forged parts, such as in this compressor disk, can be traced back to the casting process during semi-finished part production (Ill. 15.2-14). In the depicted case, a pore located near a bolt bore in the disk membrane led to cracking that spread to the surface. This crack was found with penetrant testing.
The pores are individual micropores (top left detail) or pore clusters. These pores promote cyclical crack growth in several different ways:

Pores can act as notches and increase local operating stresses considerably.

If the pores are connected with hydrogen absorption, it can result in embrittling effects with spontaneous cracking. This can even happen in new parts if the residual stresses are high enough.

If the pores are surrounded by hard brittle structure zones (“hard a), which are usually caused by oxygen or nitrogen absorption, it promotes brittle micro-cracking (Ill. 15.2-18) and increases the crack growth rate.

At the mouth of operating cracks, there is often a pore which is located in the center of a brittle, circular fracture plane (cleavage crack, quasi-cleavage crack). This fracture plane will not necessarily have crack growth lines.

Pore cracking has been observed for decades in fusion welds of titanium parts (electron beam, Ills. 16.2.1.3-30 and 16.2.1.3-31; shielding gas; Ref. 15.2-24). The cracking at the pore is largely determined by the residual stress. A hydrogen content greater than 200 ppm promotes crack initiation. An influence of oxygen has also been observed. Experts evidently agree that the brittle cracking at pores in titanium is causally related to the absorption of gases. The cracking is time-dependent, which indicates diffusion processes. It was thought that cracking had been observed during cold forming (trueing) of welds in thick-walled parts such as pressure vessels. However, this is contradicted by the expectation that these conditions would rather result in a fracture surface with typical ductile micro-characteristics (lappeting).
Today, it is evidently assumed that pore cracking in engine parts is related to dwell time fatigue (see Volume 3, Ills. 12.6.1-20, 15.2-18, and 15.2-19). This is a damage mechanism that requires a combination of static loads (constant rpm, residual stress) and cyclical loads (startup/shutdown cycles).

Example 15.2-6 (Ill. 15.2-21):

Excerpt 1 (Ref. 15.2-27): “The pilot…aborted the flight after the aircraft's No. 1 engine began emitting a thumping noise and then started to vibrate. He was able to turn the aircraft onto a high-speed access taxiway and execute a chute evacuation after receiving a report from another …aircraft that the engine was on fire.
In another incident … an aircraft (with a similar engine type) recorded a bumping noise followed by vibrations as the aircraft reached 2,000 ft. after takeoff…The pilot shut down the engine and returned the aircraft…”

Excerpt 2 (Ref. 15.2-26): ”…the safety board is recommending that the FAA require immediate, nondestructive inspection of turbine hubs…if the hubs were manufactured from Incoloy 901 and cerium or lanthanum used as oxidizing agents. Periodic inspections would be accomplished at intervals not to exceed 5,900 cycles. Investigation has revealed that a fatigue crack in the fourth-stage low-pressure turbine disk (*) extended from the hub's core to a blade slot…At a takeoff power setting, the crack expanded and allowed turbine blades to leave the disk, producing a large hole in the upper cowling…The hub had accumulated 22,022 hr. in service and 19,382 cycles that represents about 97 % of its approved service life limit of 20,000 cycles…
…the crack emanated from the bore of the turbine disk that contained `inclusions rich in cerium and lanthanum'…they rise to the top (of the ingot) as dross and are later discarded.
NTSB investigators have learned, however, that the failed hub was fabricated from a mult (forging blank) located near the top of the ingot adjacent to the dross.

Comments: (*) The description of the disk position is misleading. It actually concerns the 3rd LPT stage, but the overall 4th stage when one counts the one stage of the HP turbine.

Illustration 15.2-21 (Refs. 15.2-25 and 15.2-26, Example 15.2-6): After startup, a turbine disk made from a forged Ni alloy (Incoloy 901) fractured. It was the third low-pressure turbine stage (see engine diagram). The following inspection confirmed that an LCF crack had formed in the hub bore during operation. The crack grew radially outward to a blade groove (right diagram). A count of the crack growth lines indicated that there must already have been a crack present along the hub bore before the last overhaul. This calls into question the sufficient safety of the regulation-specified penetrant testing.
Similar cracks had already been discovered in two disks before the accident (Ill. 15.2-19, Ref.17.2-17). However, these cases did not result in disk fractures. In order to bind the oxygen in the melt, oxidizing additives of cerium and lanthanum were used in all cases. These oxides are intended to float to the top of the melt bath (Ill. 15.1-12). They collect at the top of the ingot as dross, and are cut off before further processing. However, the oxides evidently led to a segregation in the highly-stressed disk hub zone. After this was discovered, the addition of cerium was stopped. Disks that were made from raw material produced with the same methods had to be removed and non-destructively tested. The test must be repeated every 5,900 hours.
In the case of the accident disk, subsequent investigation revealed that it was located below the dross in the ingot (bottom left diagram).

Illustration 15.2-22: This engine damage occurred during the first test run of the engine, which was installed in a new aircraft. It is a single-shaft engine of an older type (middle diagram). A section of the first turbine stage disk broke out after a few minutes of total run time (bottom right diagram). The relatively small disk fragment with a blade pair broke through the turbine housing and escaped from the nacelle.
Subsequent inspection of the damaged disk revealed a large, dark oxidized flaw several square centimeters in size in the fracture zone. However, the flaw had not penetrated to the surface before the fracture. The flaw had a typical bow shape corresponding to the grain direction (Ill. 15.2-11). The surface of the residual force fracture was very small, because the flaw was located very close to the surface on both sides of the annulus.
A laboratory inspection revealed that the flaw was a large carbon nitride segregation. Experience has shown that this type of impurity can result in the disk material in question (hardenable iron-based alloy A286) during the casting process (Ill. 15.1-12).
Further research at the raw part manufacturer revealed that the disk was made from material located directly below the removed dross. This was exceptionally long in accordance with the applicable procedural controls. This indicates an increased flaw risk. Evidently, the amount of removed material was insufficient.
Macro-etching tests of the neighboring disks from the remelt ingot also revealed segregations. The greater the distance from the ingot head, the less pronounced the flaws were, and the farther inside they were located (bottom right diagram, Ill. 15.3-12). This observation can be explained with the typical flaw distribution in the ingot head (Ill. 15.3-11) and the subsequent forging process.
The prescribed penetrant test was not able to detect the internal flaw. Unfortunately, the ultrasonic testing process stipulated in the specifications was done in such a way that the the echo from the curved surface was not reflected to the receiver, which was positioned separately in this case (Ills. 17.3.1-5 and 17.3.1-9).

Illustration 15.2-23 (Ref. 15.2-28): This is a spinning disk made of a high-strength Ni-based forged alloy (Waspaloy). After vacuum induction melting (VIM), the material was remelted two times (DEVR), extruded, and heat-treated:

  • Annealed for 4 hours at 995°C - 1035°C and then quenched in oil.
  • Annealed for 4 hours at 850°C, then cooled in air.
  • Annealed for 16 hours at 760°C, then cooled in air.

The centrifugal test was done at 500°C.
The test was aborted after 3,734 cycles. A large LCF crack (3.81 x 15.24 mm) had formed from the edge of a bolt bore (left diagram). A laboratory inspection revealed that the LCF crack (bottom right diagram) spread from a 1.73 x 1.85 mm non-metallic inclusion (aluminum oxide, top right diagram).
The text does not explain how this impurity entered the material during the casting or remelting processes. However, because this is evidently a rather massive Al2O3 particle, it may have come from a ceramic filter for the melt or from lining material during melting of the ingot.

Illustration 15.2-24: A special flaw type in rolled and forged parts is two-dimensional separations that tend to run parallel to the surface. These flaws were found more frequently until the late 1960s. Today, they are evidently extremely rare. This is most likely related to improved non-destructive testing and more advanced materials technology. The separations can occur through very different mechanisms. Their shape is determined by the forming process from the cast ingot to the semi-finished/raw part. Typical flaws of this type are:

Laminations: These occur in rolled materials such as profiles and sheets. They are gas bubbles from the casting process that have been rolled flat (bottom right diagram). During heat treatment at high temperatures, especially in a vacuum, a bulge can form at the flaw. The causes are the gas pressure in the bubble and/or compressive stress caused by the more rapid expansion of the unconnected surface layer (top left diagram).
A similar effect can occur when several sheets are fused into a larger section through rolling, and local bonding problems occur. The findings in the top right detail correspond to this process. The material is a high-alloy hardenable austenitic steel (A286).

Forging laps: Occur during drop forging when material is sheared off and “smeared” by the forging motion. A typical example is precision-forged blade profiles (bottom right diagram). In this case, material from the blade root platform is transferred to the top of the blade by the forging motion or top half of the die. The depicted case concerns a compressor rotor blade made from a type of 13% Cr steel.

Separation through segregations: Larger segregations can be broken open through forging deformation and/or thermal stress. The impurity prevents the separation from resealing during subsequent forming processes. The result is internal separations that are formed into the plane of the grain direction by the forging process (Ills. 15.1-13 and 15.2-11).

Separations at unforged cracks: If cracks occur ahead of the last forging process (Ill. 15.2-11) and do not reseal, it is possible that laminar separations parallel to the surface will result. These cracks can be the result of overstressing of the material during forging. Excessive shear loads can break open the surface. Oxidation of the separation makes it impossible to fuse it shut during subsequent forging cycles. (Thermal) cracks can also occur inside forged parts. The forged part reaches a high temperature due to the applied forming energy. If this results in overtemperature and damage-promoting softening of the grain boundaries, the material can break open under the strain of the forming and/or thermal stress. However, these oxide-free separation surfaces will usually re-fuse in a subsequent forging process.
Cracks can also be caused by overly high thermal stress. If they come into contact with oxygen, resealing through subsequent forging is no longer possible.

Illustration 15.2-25: In older engine types, Cr steels were used in the compressor blading. These are heat-treated steels. The top right diagram shows a “constructed” guide vane. The blade, made from a rolled material, is soldered onto the root box, which is made of sheet metal. During operation, cracks occurred due to stress corrosion cracking (SCC, Volume 1, Chapter 5.4.2) in the root boxes (top right diagram). The cause for the susceptibility to cracking was evidently that the annealing temperature was too low during the production process (middle right diagram). The results were excessive hardness and sensitivity to SCC (bottom right diagram). At the same time, the low annealing temperature resulted in dangerously high tension residual stresses, due to the effective high yield limit (bottom left diagram).

Illustration 15.2-26 (Ref. 15.2-29): In two cases, uncontained blade fractures (right diagram) occurred in fan engines of an older type. The affected fan blades were made of a high-strength titanium alloy. Laboratory tests revealed that the fractures were HCF dynamic fatigue fractures. The fractures originated in a discolored zone below the clapper. The crack initiation zone was characterized by intercrystalline cracking and transcrystalline fracture planes. These damage symptoms are typical in this high-strength titanium alloy for stress corrosion cracking (SCC) under hot salt. Because this damage mechanism requires minimum temperatures of around 500°C (Volume 1, Ill. 5.4.2.1-8), the crack could not have occurred during operation (maximum temperature 116 °C). It was revealed that hand sweat from the blade production process was evidently the cause. The sweat was on a section of the blade that had dangerously high undetected tension residual stress from the forging process. The final heat treatment at 538 °C then caused the SCC. The cracks were evidently not detected by the non-destructive testing methods used in this case.
The left diagram shows a compressor rotor blade from an older engine design, made from a type of 13% Cr steel. The undeformed, darkly colored crack through the root shaft can probably be traced back to the forging process. It is surprising that such a flaw was not detected despite the sensitive magnetic crack detection done on the new part.

References

15.2-1 P. Adam, “Fertigungsverfahren von Turboflugtriebwerken”, Birkhäuser Verlag, 1998, ISBN 3-7643-5971-4, pages 205-206.

15.2-2 Metals Handbook Ninth Edition, “Volume 11 Failure Analysis and Prevention”, ASM,1986, ISBN 0-87170-007-7, pages 314-343.

15.2-3 “Avco Tests ALF502 Turbine Blades”, periodical Aviation Week & Space Technology”, May 31, 1982, page 14.

15.2-4 S.W. Kandebo, “GE Win Signals Entree Into F-15 Business”, periodical Aviation Week & Space Technology”, April 29, 2002, page 27.

15.2-5 R.L. Dreshfield, ” Defects in Nickel-Base Superalloys”, periodical “Journal of Metals”, July 1987, pages 16-21.

15.2-6 D. Goldschmidt, “Einkristalline Gasturbinenschaufeln aus Nickelbasis-Legierungen”, Teil I: Herstellung und Mikrogefüge. periodical “Materialwissenschaft und Werkstofftechnik”, VCH Verlagsgesellschaft, 25, 1994, pages 311-320.

15.2-7 D. Goldschmidt, “Einkristalline Gasturbinenschaufeln aus Nickelbasis-Legierungen”, Teil II: Wärmebehandlung und Eigenschaften. periodical “Materialwissenschaft und Werkstofftechnik”, VCH Verlagsgesellschaft, 25, 1994, pages 373-382.

15.2-8 D. Goldschmidt, “Single-Crystal Blades”, Proceedings der “Conference on Materials for Advanced Power Engineering”, Lüttich, Belgium, 3.Okt-6.Okt., 1994, pages 661-674.

15.2-9 H. Fredriksson, “Possible Dendrite Growth and Segregation Phenomena During Solidification of Alloy in Space””, proceedings of the “Second European Symposium on Material Science in Space”, Frascati, Italy, 6-9 April 1976, pages 291-299.

15.2-10 D. Goldschmidt, “Turbinenschaufeln aus Einkristallen”, Projekt Matfo (1560), 1991.

15.2-11 D.A. Wilson, D.P. Deluca, B.A. Cowles, M.A. Strucke, “Fatigue Crack Growth Resistance of Advanced Blade Materials”“, ASME Paper No. 86-GT-253, proceedings of the “International Gas Turbine Conference and Exhibition” Düsseldorf, Germany, June 8-12, 1986.

15.2-12 E. Fleury, L.Rémy, “Low cycle damage in nickel-base superalloy single crystals at elevated temperature””, periodical “Materials Science and Engineering”, A167, 1993 pages 23-30.

15.2-13 Metals Handbook Ninth Edition, “Volume 15 Casting”, ASM,1998, ISBN 0-87170-007-7, pages 393-430.

15.2-14 Metals Handbook Ninth Edition, “Volume 15 Casting”, ASM,1998, ISBN 0-87170-007-7,pages 546-553.

15.2-15 A.Barussaud, Y. Desvallees, J.Y. Guedou, “Control of the Microstructure in Large Titanium Discs. Application to the High Pressure Compressor of the GE90 Aeroengine”, periodical “Titanium `95: Science and Technology”, pages 1599-1608.

15.2-16 K.G.Wilkinson, “RB 211 The First Eighteen Months Operating Experience”, periodical “Tech Air”, November 1973, pages 1-9.

15.2-17 M. Nibloe, “Rolls-Royce RB.211: Der Großtriebwerkbau fordert ein hohes Lehrgeld”, periodical “Interavia” 8/1973, pages 858-858.

15.2-18 ATSB-Report No. A97F0059 1997 pages 1-29.

15.2-19 D. Hughes, M.A. Dornheim, “United DC-10 Crashes In Sioux City, Iowa”, periodical “Aviation Week & Space Technology,”, July 24, 1989.

15.2-20 “NTSB, GE Inquiry Into United DC-10 Crash Focuses On Fan Disk Recovered From Iowa Farm Field”, periodical “Aviation Week & Space Technology,”, October 16, 1989.

15.2-21 “NTSB Says CF6-6s May Require New Inspection Techniques”, periodical “Aviation Week & Space Technology”, November 13, 1989.

15.2-22 “RB.211 Investigation”, periodical “Flight International,”, 25 January, 1973.

15.2-24 T.Khaled, “An Investigation of Pore Cracking in Titanium Welds”, periodical “Journal of Materials Engineering and Performance”, Volume 3 (3) June 1994, pages 419-433.

15.2-25 “Suspect Foundry Process prompts NTSB Call for JT8D Inspections”, periodical “Aerospace Propulsion”, June 6, 1996.

15.2-26 “NTSB Targets Turbine Cracks”, periodical “Aviation Week & Space Technology”, June 10, 1996, page 30.

15.2-27 “NTSB Investigates JT8D Engine Failure”, periodical “Aviation Week & Space Technology”, June 20, 1983, page 32.

15.2-28 L.M.Jenkins, S.E.Crow, “RB211-524B Disk and Drive Cones Hot Cyclic Spinning Test”, proceedings AGARD-AR-308 of the “Propulsion and Energetics Panel Working Group 20” , September 1982.

15.2-29 E.U.Lee, R.G. Mahorter, J.D. Wacaser, “Fracture of Ti-8Al-1Mo-1V Alloy Fan Blade by Stress Corrosion Cracking and Fatigue”, ASTM Special Technical Publication 645 of the ASTM-Symposium “Fractography in Failure Analysis”,Toronto, Canada, 1-6 May 1977, pages 128-142.

15.2-30 USA FAA, “Titanium Rotating Components Review Team Report”, Volume 1, “Public Report”, December 14, 1990, Chapter 5K.

15.2-31 U.R.Kattner, “The Thermodynamic Modeling of Multicomponent Phase Equilibria”, Journal JOM ; 49 (12) (1997), pages 14-19.

© 2021 ITTM & Axel Rossmann
15/152/1522/1522.txt · Last modified: 2020/06/25 22:43 (external edit)

Page Tools